航空发动机燃油污染快速检测技术的初步研究

Preliminary Study on Rapid Detection Technology of Fuel Contamination in Lubrication System of Aeroengine

针对航空发动机润滑系统受到燃油污染的问题,提出1种基于红外光谱分析的快速检测技术。利用傅里叶变换红外光谱技术（FIRT）,结合偏最小二乘算法（PLS）,建立了燃油污染定量检测的数学校正模型。讨论了不同光谱预处理方法和PLS因子数对模型预测能力的影响。通过对光谱预处理和优化建模参数,提高了模型的预测精度,在一定的燃油质量分数范围内,得到了较为理想的数学校正模型。使用建立的分析模型对预测集样本进行预测,预测值与实际值相关性良好,相关系数R=0.9994,预测均方根误差RMSEP=0.082,重现性实验标准偏差SD=0.044~0.088。研究结果表明该燃油污染快速检测技术是可行的。

Aiming at the monitoring of contamination caused by fuel oil in lubrication system of an aeroengine, a kind of rapid detection technology base on infrared spectrum analysis technique was presented. By means of Fourier transform infrared spectrum technique （FTIR） combined with partial least square （PLS）, a mathematical calibration model for the quantitative determination of contamination caused by fuel was established. Effects of various spectral pretreatment methods and PLS factor on the prediction ability of the model were discussed. With the help of the spectral pretreatment and parameter optimization, the prediction precision of the model is improved and the acceptable mathematical model is obtained. Using the model to predict the prediction set sample, a good correlation between predictive value and actual value is found. The correlation coefficient （R） is 0.9994, the root mean square error of prediction （RMSEP） is 0.082 and the standard deviation （So） of the replicate test is 0.044-0.088. The results show that the rapid detection technology appears to be viable.